
Reputation
Badges 1
662 × Eureka!Hm, that seems less than ideal. I was hoping I could pass some CSV locations. I'll try and find a workaround for that. Thanks!
I mean, I see these are defined here https://github.com/allegroai/clearml-agent/blob/master/clearml_agent/definitions.py
But I do not see where an EnvironmentConfig.set()
is called...
I'm not too worried about the dataset appearing (or not) in the Datasets
tab. I would like it (the original task ) to to not disappear from the original project I assigned it to
No task, no dataset, just an empty container with no reference to the task it's attached.
It seems to me that it should not move the task if use_current_task=True
?
SuccessfulKoala55 That at least did not work, unless one has to specify wildcard patterns perhaps..?
~
is a bit weird since it's not part of the package (might as well let the user go through clearml-init
), but using ${PWD} works! 👍 👍
(Though I still had to add the CLEARML_API_HOST and CLEARML_WEB_HOST ofc, or define them in the clearml.conf)
I'm also getting the following warning, I guess it's some ClearML dependency?IPython could not be loaded!
Example configuration -
` version: 1
disable_existing_loggers: true
formatters:
simple:
format: '%(asctime)s %(levelname)-9s %(name)-24s: %(message)s'
filters:
brackets:
(): ccutils.logger.BracketFilter
handlers:
console:
class: ccmlp.utils.TqdmStreamHandler
level: INFO
formatter: simple
filters: [brackets]
loggers: # Set logging levels for specific packages
urllib3:
level: WARNING
matplotlib:
level: WARNING
...
Couldn't the agent just come with the toml library? Kinda easy to load up and check if poetry is present then... 🤔
But yes it indeed used poetry correctly, though it would fail in other circumstances
Some examples of the mess it creates (also posted in the main channel):
A single project now has multiple subprojects The subprojects have the .datasets
hidden subproject (with really frustrating project names) The subprojects are empty To access the original project, I have to go twice into the same project because of these hidden projects Because of these hidden subprojects, I cannot delete a project that has 0 experiments
SuccessfulKoala55 help me out here 🙂
It seems all the changes I make in the AWS autoscaler apply directly to the virtual environment set for the autoscaler, but nothing from that propagates down to the launched instances.
So e.g. the autoscaler environment has poetry
installed, but then the instance fails because it does not have it available?
Oh! Nice! I'll have a go at it and report back at the PR if it's in a functional state 🙂 Thanks AgitatedDove14 !
I’ll also post this on the main channel -->
AgitatedDove14
I'll make a PR for it now, but the long story is that you have the full log, but the virtualenv
version is not logged anywhere (the usual output from virtualenv
just says which Python version is used, etc).
I wouldn't put past ClearML automation (a lot of stuff depend on certain suffixes), but I don't think that's the case here hmm
Indeed with ~
the .root
call ends with an empty string, so it has a bit of different flow
This could be relevant SuccessfulKoala55 ; might entail some serious bug in ClearML multiprocessing too - https://stackoverflow.com/questions/45665991/multiprocessing-returns-too-many-open-files-but-using-with-as-fixes-it-wh
i.e.ERROR Fetching experiments failed. Reason: Backend timeout (600s)
ERROR Fetching experiments failed. Reason: Invalid project ID
I see, okay that already clarifies some stuff, I'll dig a bit more into this then! Thanks!
So basically I'm wondering if it's possible to add some kind of small hierarchy in the artifacts, be it sections, groupings, tabs, folders, whatever.
I guess the thing that's missing from offline execution is being able to load an offline task without uploading it to the backend.
Or is that functionality provided by setting offline mode and then importing an offline task?
Or do you mean the contents of the configuration, probably :face_palm: ... one moment
Then perhaps mac treats missing environment variables as empty and linux just crashes? Anyway, the config loading should be deferred, shouldn't it?
I'm guessing that's not on pypi yet?
Here's a full description of the layout:
Remote agent + entire ClearML docker suite running on host A. Host A also has a /data/clearml
folder mounted to it and to it's docker containers (I've edited the docker-compose
to add this mount point) Connect to host A, use StorageManager on the /data/clearml
folder for some early troubles (e.g. long .list
call) Use the same connection to run a task with execute_remotely
and download_folder
and see it crash :disapp...
An internal project I've accidentally made with a hidden tag while playing around with the ClearML internal code.
We just do task.close() and then start a new task.Init() manually, so our "pipelines" are self-controlled
No it doesn't, the agent has its own clearml.conf file.
I'm not too familiar with clearml on docker, but I do remember there are config options to pass some environment variables to docker.
You can then set your environment variables in any way you'd like before the container starts